InRhythm

Your partners in accelerated digital transformation

Code Lounge

Why you should migrate away from Angular JS

Hello readers! A big fan of the Angular Javascript framework here. Today I wanted to go through a few of the reasons why you should look to migrate your existing AngularJS projects (any Angular release version under 2) to a more modern and actively supported framework (or library).

You might be thinking “well, this guy is internally conflicted”, but give me the opportunity to explain myself. Even though AngularJS is a fantastic piece of technology that surely was top of its class when it came out (October 2010) and despite the fact that I enjoy working with its successor Angular.io (also known as Angular 2+), AngularJS has become outdated (EOL December 2021), and a risk to your company in many different ways.

Dilbert strip - Larry tries to tell PHB what the legacy systems did. PHB says no, he doesn't care about legacy systems. Dilbert asks, "I though we replaced all of our legacy systems." Larry says, "Keep your thoughts to yourself."
@2017, Scott Adams, Inc.

The framework has reached end of support (Version Support Status). This means that it has become read-only mode, it will not be updated further. The framework has not been developed for over a year now (Release 1.8.2 happened in October 2020), and even though extended support was supposed to end mid-2021, it was extended to December 2021 due to the global pandemic. Here’s a blog post by the Angular team regarding discontinued long term support.

To add some support to the point I’m trying to make, I’ll share that Angular was created and mainly maintained by Google. Google recognized the shortcomings of AngularJS, and completely rewrote it to release Angular.io. AngularJS only made it to version 1.8.3, however Angular.io has already made it to major version 13 (current at time of writing), with many more versions to come.

What could be making you hold onto your existing AngularJS apps

Trust me, I’ve been there. You have a perfectly functioning application which needs little maintenance, and you have engineers who know it in and out already. Why invest a part of your budget in fixing something that’s not broken? Why bring in new people that don’t know the product? Why push your engineers to do something new/different to what they’ve been doing?

Dilbert strip - PHB announces to Dilbert that the legacy system maintainer quit. Tells Dilbert to fix it. Dilbert standing in front of odd large mechanical device with a screen, pipes, and pipes with the thought bubble, "Frack."
@2016, Scott Adams, Inc.

The reasons

Technology: As I stated earlier, AngularJS is outdated. This means that in its feature-set, performance, and just keeping up with latest developments in Javascript and the web browsers, AngularJS has clearly lagged behind, mainly due to the fact that it has been in maintenance mode and not actively developed on for years. If you stay on this framework, you won’t take advantage of the rapidly evolving web world, and the evolving smart devices and their new features.

Support: As the framework is no longer maintained, any new issues or limitations you encounter will not only lack an answer/help from the AngularJS team (again, not supported anymore), you most probably also won’t have a huge community online to help you with it, like you would have with any modern framework. This could mean a longer time to fix issues that come up in your application, and a rough experience for your engineers and users.

Security: Perhaps the biggest reason why you should move away from AngularJS. Like any unsupported package out there, you won’t be protected when any new security exploits are identified, be it within the framework itself, or any of its thousands of dependencies and indirect dependencies (yes, your app can be exploited by vulnerabilities in the dependencies of the dependencies of AngularJS which is your app’s dependency… you get the point). Usually when something like this happens in an actively supported package, a fix will be published quite swiftly in response to it, or any dependency that includes the vulnerability will be updated with a newer version.

Talent: Not only do you want to provide the best possible experience for your users, but also for your app engineers. When you are trying to retain or expand your software team, AngularJS will weigh on any engineer’s decision. Engineers will want to work with quality, cutting edge technology. It is hard for engineers to get or even stay excited about working on a framework that has reached end of life. It will be much easier for you to retain and hire engineers if your apps run on modern technologies and following best practices and industry trends. I cannot stress how much easier will it be to fill open positions when your tech stack is attractive for the engineers. You can also think about what will happen once you actually find someone willing to do the job on your legacy system, they will play hard to get and you’ll end up paying more for an engineer that is probably not up to date on industry standards.

Business: For current technologies, the help you will get from the online community is massive, which speeds up the time it takes to fix and implement new features, and to resolve critical situations that may arise. Not only your engineers will be happier and more engaged in what they are doing, it also impacts your branding. Are you a company that invests in and works with the latest and greatest? Or a company that settles with whatever is there?

Dilbert strip - Patty asks Larry how long it will take to add a feature to the legacy system. Larry asks when the new system will replace the old system. Patty responds, six months. Larry then says the new feature will take seven months.
@2017, Scott Adams, Inc.

I know that was a lot of talking, but I can tell you with confidence that we have seen the impact that migrating legacy applications has for many of our customers, and it is massive. Not only do applications come alive and look and feel more modern, but engineers come to work in a better mood and eager to get things done, and a true engineering culture is fostered. Even though keeping legacy systems might seem like the easy or cost-efficient way, there are several hidden (or not obvious) costs that come with it.

If you need any help assessing or migrating your systems, do not hesitate to reach out to us at any of the following:

get@inrhythm.com

InRhythm Website – Contact Us

+1 (800) 683-7813

Written by Juan Porley, Director of Engineering, Web Practice @ InRhythm

SwiftPM – The Swift Package Manager

Swift Package Manager

Introduction

The Swift Package Manager (SwiftPM) is Apple’s tool for managing package dependencies for Swift application development. SwiftPM has been an integrated part of Swift since v3.0.

So, what are Packages? Packages contain reusable code or other resources stored in repositories that your application needs to provide a feature or function. Some examples include, but are not limited to, the fonts and color scheme for your app, a library that provides access to a web resource, a file with images required by your application. Other examples are the APIs and Frameworks that add functionality to your app and simplify your coding tasks.

This post will provide a high-level overview of basic SwiftPM usage and features.

What do Package Managers bring to the Table?

Package Managers give developers a way to control which packages and which versions of those packages that the project will consume.

Other features include:

  • Allows easy use of light-weight, reusable libraries
  • Reduces code duplication
  • Supports development best practices by simplifying and supporting modular coding (easy module/package creation)

While it is possible to manually manage dependencies in a project, it isn’t practical for anything but the smallest of applications. Best practices fall on the side of using a package manager to handle this function automatically.

SwiftPM supports Swift 3+ and XCode 8+. Since Swift 5 and Xcode 11, SwiftPM has had cross-platform support for iOS, macOS, and tvOS.

Are there SwiftPM Alternatives?

There are two mainstream alternatives, CocoaPods and Carthage. Both are third-party tools requiring installation and much configuration before use.

CocoaPods

CocoaPods (2011) – Supports Swift 5+ and Xcode 3.11+ – Mature and stable. A centralized, easy-to-use command-line tool for package dependency management and integration into the application. CocoaPods is written in Ruby.

However, Cocoapods lacks direct control over project configuration and is basically a blackbox. CocoaPods also experiences random build failures that can’t be explained. Resolving these often involves removing and reinstalling CocoaPod dependencies from the project, clearing caches, and completely rebuilding the entire project. These steps can add a significant amount of time to the development process, especially on large projects.

Each CocoaPod dependency must have a “Podspec” file to define the metadata for that dependency. These files are typically uploaded to a different repository from the dependency itself. A “Podfile” within the app project will include a reference to this repository, and CocoaPods will use the information in this Podfile to fetch and install all of the dependencies for the app project. 

Carthage

Carthage (2014) – A decentralized command-line tool for dependency management. Developers have full project control and must manually provide all package and dependency information. Carthage pulls the packages the developer specifies.Configuration requires much more manual work than CocoaPods and is somewhat complicated.

Using SwiftPM

SwiftPM has many advantages over Carthage and CocoaPods. Foremost, SwiftPM is a native Apple tool integrated into Swift. Other advantages include a quick and simple configuration, easier control over packages and their sub-dependencies, and a GUI built into Xcode for managing package configurations. Each package’s metadata is defined in a “Package.swift” file, which  resides in the same repository as the package source files. 

Package Resolution

With the Package dependencies configured in the SwiftPM GUI, Xcode will download the packages and resolve dependencies at build time with no further developer interaction.

When setting dependencies, you have the following options:

  • By Version: Next Version, Up to next minor, Range, or a specific commit
  • Branch (Specify)
  • Commit(Specify)

Adding a Package to your Project

  1. Swift Packages> Add Package Dependency 
  2. On the “Choose Package Repository” dialog, enter the desired Repository.
  3. Click Next. 
  4. On the “Choose Package Options” dialog, set the dependency rules (Versions, Branches, Commits). 
  5. Click Next. Xcode now fetches the dependency.
  6. On the “Add Package to YourPrjName dialog,” ensure that the relevant packages are checked and the proper Target is selected in the Add to Target cell.
  7. Click Finish.

Your package now appears in the Navigator under Swift Package Dependencies. Note that SwiftPM can reference both local and remote packages.

Create a Module and Add a Local Package with SwiftPM

SwiftPM simplifies the process of modularizing your code and adding it as a package to a local or external repository. This section will briefly describe the process of creating a module and adding it to a local repository.

  1. Identify a self-contained portion of code or another resource that is suitable for use in a module.
  2. Create a new file: Files > New Package. Name the package accordingly and add it to your application using the same dialog.
  3. Copy the code or resource to the new file.
  4. Delete the existing code or resource from the app
  5. Add the new package to the Application Target. In “Application Project Settings,” add the new package under Frameworks and Libraries.
  6. In the Package code, define the platforms and versions where this Package works.
  7. In the Application’s code files, import the new package into every file where its code or where its resources are used. 

If desired, you could upload the new Package to an external repository for use by developers outside your organization.

Additionally, SwiftPM will allow you to create and use binary packages. Binary packages permit developers to distribute libraries and frameworks without also distributing their source code.

Package Distribution

You can use SwiftPM to distribute resources, in addition to frameworks and libraries. SwiftPM also has built-in support for Apple’s  DocC Documentation format, which makes it easy to build robust interactive documentation files and tutorials. 

The Future

SwiftPM is a mature product still under active development to improve and add new features. Usage stats currently show it being about equal in popularity with CocoaPods. However, since SwiftPM is a native tool that requires no additional work to begin using, its future is more sure than that of CocoaPods or Carthage.

Resources

InRhythm University Lightning Talk: Lightning Talk / Swift Package Manager / January 27th 2022 / Andrew Balmer

SwiftPM at Swift.org: https://www.swift.org/package-manager/

SwiftPM Repository: https://github.com/apple/swift-package-manager

State Management: Redux vs. MobX Comparison

Redux versus MobX

This post discusses two State Management packages. The first library, Redux, was developed in 2015 by Abramhov and Clark based on concepts used in Facebook’s Flux architecture. MobX also began in 2015 and was originally called Mooservable by the dev team. It became MobX with the 2.0 release on 26 Feb 2016.

There are three parts to state management. The State, which is the source of truth for your application’s State; the View, which is a rendering of the current State; and the Action, which is something that triggers a State change.

Bare Javascript passes State via property trees. A function needing State obtains the information via property drilling. Depending on a function’s layer depth, property drilling can introduce significant overhead.

Using State Management introduces a structured workflow in the application and makes obtaining the application State trivial.

Redux

Redux Overview

Redux manages application state in a centralized repository. There is a single state in the application and it is immutable. It can’t be modified by the application. This is similar to how State is managed in Flux. Redux is designed to be a predictable container for application state.

Redux combines Flux and Functional Programming and uses an immutable State as a single, predictable, source of truth in the application stored in a Javascript object. 

Redux Implementation

Using Redux requires a considerable amount of ramp-up work. There is a great deal of boilerplate code, configuration, and to take advantage of some capabilities, developers must follow explicit coding patterns. 

Redux Workflow

There are three main workflow components:

  • The UI
  • The Event or Action Handler/Dispatcher
  • The Store with Reducers and the State

The UI passes the user’s activity, or their Action, to the Event Handler. The Event Handler’s Dispatcher then notifies the Store of the Action. The Store clones a copy of the current State for the Reducers to act on based on the user’s Action. On completion,the Store passes the new state to the UI. The updated State will trigger a refresh of any component accessing that State.

The State’s Reducers are responsible for updating the State of the Store. They are pure functions that analyze the Store’s recursive data structure. They do not mutate the current state, nor do they make external API calls. Using a combining function, they recombine the results.of their recursive processing into their constituent parts, thereby building a return value.

MobX

MobX Overview

MobX was created in ES5 Javascript and follows Functional Reactive Programming patterns and prevents an inconsistent State by automatically performing all derivations. MobX’s main difference from Javascript and Redux are that State is not immutable and your application can automatically store different observable data types into different MobX Stores.

MobX uses abstraction to hide much of the background operation and reduce the boilerplate code requirement. MobX directly modifies its current state(s) without cloning beforehand.

MobX Implementation

MobX is implicit by nature, and requires little boilerplate code. After implementing State storage in the desired data structure (plain, arrays, objects, etc.), set those properties that will change over time to observable for MobX tracking.

MobX Workflow

There are four main workflow components:

  • Actions
  • Observable State
  • Derivations (Computed Values) 
  • Reactions (or Side Effects)

Events invoke Actions. Those actions can originate with the user or the application, and are the only thing that can modify the Observable State. MobX propagates any changes to the Observable State to all Derivations and Reactions. 

Deciding on a State Manager

Types of Derivations could be computation, serialization,  rendering, etc. and example Rendering side-effects are I/O activity, DOM updating, network activity. etc.

  • Application Architecture – Is the architecture Flux/Functional Reactive Programming (Redux) or more OOP (MobX)?
  • Ramp-up – Time until productive
    • Boilerplate
      • Redux requires extensive boilerplate setup
      • MobX has very little boilerplate
    • Learning Curve
      • Steeper – Redux requires developers to follow explicit coding patterns to access capabilities. 
      • Flatter – MobX is more implicit and doesn’t require extra tooling. Setup is adding a class and setting Observers, Actions, and Compute Values.
  • State Stores – Redux usually maintains a single immutable State even though it is possible to set up multiple Stores in Redux. It is a cumbersome process. MobX makes it much simpler to have multiple Stores and it uses changeable States.
  • Purity
    • Redux uses pure functions and does not allow State modification.
    • MobX allows mutable States. 
  • Scalability
    • Redux’s pure functions make code scaling much simpler and predictable. 
    • MobX’sMobX’s workflow does not include many pure functions, making testing not as straightforward, and as the application grows it becomes more difficult to determine what exactly should an output be, given an input.
  • Performance
    • Redux can suffer performance degradation from very frequent updates to its single State Store. 
    • MobX, on the other hand, with support for multiple Stores can better support frequent State updates, such as real-time updates on market pricing for a trading platform.
  • Community Support and Tools
    • Redux has a much larger community and more widely tested tools.
    • MobX has a smaller community and fewer tools.
  • Debugging
    • Redux has a single source of truth and uses a pure Store making its output predictable and repeatable. Redux has better debugging tools, including ReduxDevTools and the Time Traveling Debugger (provides State history).
    • MobX with its implicit nature, multiple States, and code abstraction can be very difficult to troubleshoot and determine the proper output from a given input. All State updates happen behind the scenes and do not offer any history. MobX also suffers from fewer robust tools.

Resources

InRhythm University Lightning Talk: Lightning Talk / State Management: Redux vs MobX / February 17th 2022 / Eric Marcatoma

Redux: https://redux.js.org/

MobX: https://mobx.js.org/the-gist-of-mobx.html

Creating Robust Test Automation for Microservices

Any and all projects that a software engineer joins will come in one of two forms: greenfield or legacy codebases. In the majority of cases, projects will fall into the realm of the legacy repositories. As a software engineer, it is our responsibility to be able to strategically navigate our way through either type of project by looking objectively at the opportunities to improve the code base, lower the cognitive load for software engineering and make a determination to advise on better design strategies. But, chances are, there is a problem. Before architecture or design refactors can be taken its best to take a pulse on the health of your platform End to End (E2E). The reason being, lurking in a new or existing platform is likely a common ailment of a modern microservices approach – the inability to test the platform E2E across microservices that are, by design, commonly engineered by different teams over time.

I recently worked on a greenfield platform and nearly 6 months into the project, one of the most critical parts of the project had fallen several months behind from a quality assurance perspective. It was no longer possible for QA to catch up, nor was it possible for QA to engineer and execute E2E testing to complete common user journeys throughout the enterprise system. To solve this conundrum, E2E data generation tools needed to be created so that the QA team could keep upbuilding and testing every scenario and edge case. As I see it, there are three main requirements for an E2E account and data generation tool. The tool should:

1) Create test accounts with mock data for each microservice

2) Link those accounts between up and downs stream microservices

3) Provide easy to access APIs that are self-documenting 

Using a tool like Swagger, QA can use the API description for REST API, i.e. OpenAPI Specification (formerly Swagger Specification) to view the available endpoints and operations to create accounts, generate test data, authenticate, authorize and “connect the microservices”.

By creating tools for E2E testing, our QA team was able to eliminate the hassle of trying to figure out which upstream and downstream microservices needed to be called to ensure that the required accounts and data were available and set up properly to ensure a successful test of all scenarios i.e. based upon the variety of different data types, user permissions, user information and also covering the negative test cases. The QA team was able to catch up and write their entire suite of test scenarios generating the matching accounts and data to satisfy those requirements. The net result of having built an E2E test generation tool was automated tests could be produced exponentially quicker and the tests themselves are more resilient to failure. 

Even though the microservices pattern continues to gain traction, developing E2E testing tools that generate accounts and test data across an enterprise platform will likely still remain a pain point. That having been said, the purpose of this article is to raise awareness. Software Engineering Leadership has an opportunity to step up and ensure that the enterprise system that they look after is healthy E2E. I can’t think of a better way to maintain a healthy system than to ensure accounts and data in the lower environments actually work and unblock testing end-to-end. 

James Woods, Web Engineering Practice Lead